| description abstract | A coupled air?sea numerical model comprising a mesoscale atmospheric model, a marine circulation model, and a surface wave model is presented. The coupled model is tested through simulations of an event of frontal passage through the Lake Erie region. Experiments investigate the effects of different sea surface roughness parameterizations on the atmospheric simulations. The coupled system?s components are the fifth-generation Pennsylvania State University?National Center for Atmospheric Research Mesoscale Model (MM5), the Princeton Numerical Ocean Model (POM), and the GLERL?Donelan Wave Model (GDM). The finest of the MM5?s three nested grids covers Lake Erie, on which the POM and GDM operate. The MM5 provides surface heat and momentum fluxes to the POM, and the POM returns lake surface temperatures to the MM5. The MM5 provides 10-m winds to the GDM, and the GDM returns sea state information to the MM5. The MM5 uses this sea state information in calculating overwater roughness lengths (z0?s). Experiments varying the MM5?s roughness parameterization over Lake Erie are performed, resulting in a broad range of z0?s. It is found that wave model coupling can significantly increase overwater roughnesses in the MM5, leading to increased surface heat and moisture fluxes and to changes in PBL characteristics. The impacts on the atmosphere from marine model coupling can appear far downstream of the coupled zones. The accuracy of the mesoscale atmospheric simulation appears sensitive to the assumptions behind the marine roughness parameterizations used. The results suggest that, for consistent forecast improvement, marine roughness parameterizations should account for wave age. In addition, it is found that accounting for wave movement in an air?sea coupling scheme can be a significant factor in the calculation of surface stresses and, with them, surface heat fluxes over marine areas. Thus, the approach with which a coupling scheme implements sea-state-dependent roughness parameterizations can be as influential as the parameterizations themselves. | |
